Professional Documents
Culture Documents
Textbook Strong Fermion Interactions in Fractional Quantum Hall States Correlation Functions Shashikant Mulay Ebook All Chapter PDF
Textbook Strong Fermion Interactions in Fractional Quantum Hall States Correlation Functions Shashikant Mulay Ebook All Chapter PDF
Textbook Strong Fermion Interactions in Fractional Quantum Hall States Correlation Functions Shashikant Mulay Ebook All Chapter PDF
https://textbookfull.com/product/strong-interactions-in-
spacelike-and-timelike-domains-dispersive-approach-alexander-v-
nesterenko-auth/
https://textbookfull.com/product/electron-correlation-in-
molecules-ab-initio-beyond-gaussian-quantum-chemistry-1st-
edition-hoggan/
https://textbookfull.com/product/geometry-of-quantum-states-an-
introduction-to-quantum-entanglement-ingemar-bengtsson/
https://textbookfull.com/product/geometry-of-quantum-states-an-
introduction-to-quantum-entanglement-ingemar-bengtsson-2/
Excursions in Ill Condensed Quantum Matter From
Amorphous Topological Insulators to Fractional Spins
Adhip Agarwala
https://textbookfull.com/product/excursions-in-ill-condensed-
quantum-matter-from-amorphous-topological-insulators-to-
fractional-spins-adhip-agarwala/
https://textbookfull.com/product/a-new-platform-for-edge-mode-
manipulations-in-the-quantum-hall-effect-yonatan-cohen/
https://textbookfull.com/product/ideas-of-quantum-chemistry-
volume-2-interactions-3rd-edition-lucjan-piela/
https://textbookfull.com/product/quantum-physics-states-
observables-and-their-time-evolution-arno-bohm/
https://textbookfull.com/product/plant-microbe-interactions-in-
agro-ecological-perspectives-volume-1-fundamental-mechanisms-
methods-and-functions-1st-edition-dhananjaya-pratap-singh/
Springer Series in Solid-State Sciences 193
Shashikant Mulay
John J. Quinn
Mark Shattuck
Strong Fermion
Interactions in
Fractional Quantum
Hall States
Correlation Functions
Springer Series in Solid-State Sciences
Volume 193
Series editors
Bernhard Keimer, Stuttgart, Germany
Roberto Merlin, Ann Arbor, MI, USA
Hans-Joachim Queisser, Stuttgart, Germany
Klaus von Klitzing, Stuttgart, Germany
The Springer Series in Solid-State Sciences consists of fundamental scientific books
prepared by leading researchers in the field. They strive to communicate, in a
systematic and comprehensive way, the basic principles as well as new
developments in theoretical and experimental solid-state physics.
Mark Shattuck
123
Shashikant Mulay Mark Shattuck
Department of Mathematics Department of Mathematics
University of Tennessee University of Tennessee
Knoxville, TN, USA Knoxville, TN, USA
This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface
v
vi Preface
section of Chap. 2 make a novel contribution to invariant theory, and hence, they
are of interest from a purely mathematical standpoint. Our desire to present this
attractive symbiosis between the theory of invariants and the correlation polyno-
mials G serves as one of the motivations for this monograph. The first Appendix A
is devoted to proving that our intuitive approach is indeed applicable to the
Moore-Read state and essentially reproduces the known trial wave function. In the
second Appendix B, we pose some currently unresolved problems closely related to
the mathematics in Chap. 2 and brought to light by our investigations [3, 4]. The
last appendix records some light computational procedures that we have used. We
have striven to make this monograph as broadly accessible as possible so that it can
be used as a text in an advanced course dealing with the fractional quantum Hall
effect.
References
1. S.B. Mulay, J.J. Quinn, M.A. Shattuck, Correlation diagrams: an intuitive approach to corre-
lations in quantum Hall systems. J. Phys: Conf. Ser. 702, 1–10 (2016)
2. S.B. Mulay, J.J. Quinn, M.A. Shattuck, A generalized polynomial identity arising from
quantum mechanics. Appl. Appl. Math. 11, 576–584 (2016)
3. S.B. Mulay, J.J. Quinn, M.A. Shattuck, An algebraic approach to electron interactions in
quantum Hall systems. arXiv:1809.01504 (2018)
4. S.B. Mulay, J.J. Quinn, M.A. Shattuck, An algebraic approach to FQHE variational wave
functions. arXiv:1808.10284 (2018)
Acknowledgements
First and foremost, we are very grateful to Luis Finotti for his frequent and generous
aid in carrying out computations via SAGE as well as MAPLE; his expert help was
essential in computing the third row of the last column of Table 2.2. We thank
George Simion for doing some numerical testing for us. We are grateful to Brendan
McKay for his advice on the gtools software [1] which allowed us to compute the
entries of the second row of Table 2.2. Most of the correlation diagrams of this
monograph were drawn using the multigraph package [2] of the R library; we are
grateful to Antonio Rivero Ostoic for answering our queries about this package. We
thank Abdelmalek Abdesselam and Jaydeep Chipalkatti for providing the reference
[3]. Lastly, we thank Marie Jameson for providing a MAPLE procedure which lists
the sets EðN; dÞ (for arbitrary N and d).
References
1. B.D. McKay, A. Piperno, Practical graph isomorphism, {II}. J. Symbolic Comput. 60(0),
94–112 (2014)
2. A.R. Ostoic, multigraph: plot and manipulate multigraphs (2017). R package version 0.75
3. J. Dixmier, Quelques résultats et conjectures concernant les séries de Poincaré des invariants
des formes binaires, in Séminaire d’algébre Paul Dubreil et Marie-Paule Malliavin, 36éme
année (Paris, 1983–1984), Lect. Notes Math., vol. 1146 (Springer, Berlin, 1985), pp. 127–160
vii
Contents
ix
Symbols
DðzÞ, 30
dðz; AÞ, 24
jG , 27, 88
”, 3
…ðBÞ, 32
…½N, 33
vðz; C; eÞ, 33
bound ðE Þ, 69
CF, 2, 3, 5–11, 13–15, 17–20, 22
CS, 4, 5, 8, 17, 18
dinvk ðN; d Þ, 27
Dðm;nÞ , 30
Dn , 31
E ðN Þ, 24
E ðN; V Þ, 24
E ðN; V Þ, 24
E ðN; d Þ, 24
E ðN; d Þ, 24
frqðb; EÞ, 74
FQH, 4
GðN; d; XÞ, 27
GL , 88
GLðn; kÞ, 10
Hk ðm; NÞ, 15
Invk ðN; d Þ, 27
IQL, 2–5, 7–13, 15, 18, 19, 21, 22
LL, 2, 6, 14
maxð AÞ, 46
M ðm; n; a; cÞ, 33
M ðn; a; cÞ, 34
xi
xii Symbols
M0 ðm; n; a; cÞ, 31
MF, 5, 11, 17
MFCF, 2, 3, 6, 7, 11, 17–19, 111
Mðr; sÞ, 61
M þ ðr; sÞ, 61
M2 ðr; sÞ, 61
pðm; d; nÞ, 16
QE, 2, 3, 6, 8–13, 18–20
QH, 2, 3, 6, 9, 13, 18
QP, 2, 4, 6–9, 18
RPA, 2
symgrpðEÞ, 70
SN , 11
SymgrpðEÞ, 70
SymmN , 11
SLðn; k Þ, 12
Chapter 1
Fermion Correlations
1.1 Introduction
Solid state theory developed from Sommerfeld’s realization [1] that simple metals
could be described in terms of a gas of free quantum mechanical electrons that
obeyed the Pauli exclusion principle [2]. The electrical and thermal conductivities,
heat capacity, spin susceptibility and compressibility predicted by this free electron
model agreed with experimental observations in simple metals like Al, Na, and K.
Early work on the effect of the periodic potential of the solid on the single electron
eigenstates [3] led to the concept of energy bands and band gaps, and to some
understanding of why some solids were metals, some were insulators, and others were
semiconductors [4]. During the early decades of solid state theory, the description
of the electronic states rested on the “single particle” picture.
In the middle of the last century, scientists began to question why Sommer-
feld’s simple model worked so well. The model completely neglected the strong
−1
Coulomb interaction V (r) = i< j e2 r i − r j of the electrons with one another.
Treating V (r) in first orderperturbation
theory added an exchange energy Ex (k) =
k 2 −k 2
− e2 k F /2π 2 + Fkk F ln kk FF +k−k
to the free electron kinetic energy E(k) =
2 k 2 /2m. In these equations, k F is the Fermi wave number, and it is given by
1/3
k F = 3π 2 n 0 , where n 0 is the electron density. The exchange energy ruined
the agreement with some of the experimental results. In addition, going beyond
first order perturbation theory led to divergences. Feynman diagrams depicting the
Coulomb interaction to any order in perturbation theory were used in evaluating the
ground state energy [5]. Summing certain sets of diagrams to infinite order before
integrating over the wave vector removed the divergences that resulted from the long
range of the Coulomb interaction. Several authors [6, 7] emphasized the polarization
of the electron gas by a moving charged particle, following up on the pioneering
work of Lindhard [8, 9]. The “self-energy ”of a single excited electron interacting
with the polarization cloud that it induces around itself was evaluated [6]. The effec-
set of fractional quantum Hall states observed experimentally. Chen and Quinn [19]
introduced an “effective CF angular momentum” l0∗ = l − p(N − 1) associated with
the lowest CF Landau level (CFLL0). For N = 2l0∗ + 1, this level is exactly filled
and a Jain IQL state results. If N > 2l0∗ + 1, then N − (2l0∗ + 1) particles must be
placed in the next angular momentum shell with l1∗ = l0∗ + 1; these are CFQEs. If
N < 2l0∗ + 1, there will be 2l0∗ + 1 − N CFQHs in CFLL0. Thus we have lQH = l0∗
and lQE = l1∗ = l0∗ + 1. For any given value of l, the single electron angular momen-
tum, one can obtain the number of QEs in the partially filled shell (or the number
of QHs in the partially unfilled shell). The lowest band of angular momentum states
will contain the minimum number of CFQP excitations consistent with the values of
2l and N . The value of (2l, N ) defines the function space of the N electron system.
resulted because all of the electron pairs were able to avoid pair states with relative
angular momentum R2 (R2 ≡ 2l − L 2 ) smaller than m (or separation smaller than
m 1/2 λ). The avoided pair states had the largest Coulomb repulsion, and avoiding all
such states should give an energy minimum. Laughlin proposed a many-body wave
function at ν = m −1 given by
− |z k |2
m (1, 2, . . . , N ) = exp k
z i j |m| . (1.1)
4λ2 i< j
For m = 1, the product i< j z i j (where z i = ri e+iφi ) is just the Fermion factor which
|m−1|
keeps the non-interacting electrons apart. The remaining factor, G = i< j z i j , is
a symmetric correlation factor caused by the Coulomb interactions. It is not difficult
to see that ν = m −1 , so that m = 1 corresponds to a filled LL0, and that m = 3
corresponds to a one third filled level. Laughlin also showed that the elementary
excitations of the IQL state could be described as fractionally charged QEs and QHs.
Both localized and extended states of the quasiparticles were required to understand
the behavior of ρx x and ρx y .
The first explanation of the FQH states at filling factors ν = n (1 + 2 pn)−1 with
n > 1 was given by Haldane [23, 24]. He assumed that the dominant interaction
between quasiparticles was the short range part of the pair interaction. If this interac-
tion were sufficiently similar to the Coulomb interaction in electron LL0, the problem
of putting NQP QPs into a QP Landau level would be essentially the same as the origi-
nal problem of putting N electrons into electron LL0. The number of QP states in the
QP Landau level could not exceed N , the original number of electrons. This led Hal-
dane to the condition N = 2 pNQP in place of Laughlin’s condition Nφ = (2 p + 1) N
for the electron IQL states. He picked an even integer 2 p in place of Laughlin’s odd
2 p + 1 because he considers the QPs to be Bosons instead of Fermions. Haldane’s
hierarchy of IQL states contained all odd denominator fractional filling.
Jain [16, 17] introduced a simple composite Fermion picture by attaching to each
electron (via a gauge transformation) a flux tube which carried an even number, 2 p,
of magnetic flux quanta. This “Chern–Simons”(CS) flux [25] has no effect on the
classical equations of motion since the CS magnetic field b (r) = 2 pφ0 i δ (r − r i )
vanishes at the position r i of each electron (it is assumed that no electron senses its
own Chern–Simons flux). Here φ0 is the quantum of flux, and the sum is over all
electron coordinates r i . This CS transformation results in a much more complicated
many-body Hamiltonian which includes a CS vector potential a (r) given by
ẑ × (r − ri ) †
a (r) = αφ0 d 2 ri ψ (r i ) ψ (r i ) (1.2)
(r − ri )2
1.3 Jain’s Composite Fermion Approach 5
Table 1.1 The effective CF Monopole strength 2Q ∗ , the number of CF quasiparticles (quasiholes
−n QH and quasielectrons n QE ), the quasiparticle angular momenta lQH , lQE and the angular momenta
L of the lowest lying band of multiplets for a ten electron system at 2Q from 25 to 29
2Q 29 28 27 26 25
2Q ∗ 11 10 9 8 7
n QH 2 1 0 0 0
n QE 0 0 0 1 2
lQH 5.5 5 4.5 4 3.5
lQE 6.5 6 5.5 5 4.5
L 10, 8, 6, 4, 2, 0 5 0 5 8, 6, 4, 2, 0
electrons when 2Q ranges from 25 to 29. The total angular momentum L is obtained
by addition of Q P angular momenta of n QP quasiparticles treated as Fermions.
In the mean field approximation, the CFQPs do not interact with one another.
Therefore, the states with two QEs (L = 0 ⊕ 2 ⊕ 4 ⊕ 6 ⊕ 8) and with two QHs (L =
0 ⊕ 2 ⊕ 4 ⊕ 6 ⊕ 8 ⊕ 10) should form degenerate bands. Numerical diagonalization
of the ten electron system clearly shows that the two quasiparticles states are not
degenerate. The deviation of the energies E (L) from their non-interacting energy
E 0 = 2EQP gives their pair interaction energy as a function of their pair angular
momentum L 2 up to a constant which does not influence correlations. Figure 1.1
shows E (L), the energy as a function of total angular momentum for the ten electron
system. It is apparent that Jain’s MFCF picture gives the correct values of L for
the lowest band of states obtained by exact numerical diagonalization (within the
subspace of the partially filled LL).
For large systems (e.g., N >14), numerical diagonalization of the electron-electron
interactions becomes difficult, so we have investigated the low lying energy states
by determining the number of QEs and QHs (n QE or n QH ), their angular momenta
lQE and lQH , and their interaction energies VQE (L 2 ) and VQH (L 2 ). Since n QE (or
n QH ) is much smaller than N , and lQE (and lQH ) much smaller than l, the electron
angular momentum, we can easily diagonalize these smaller systems. One exam-
ple [18] is shown in Fig. 1.2 for the case (2l, N ) = (29, 12), which corresponds to
2lQE , n QE = (9, 4). The low lying states of the electron system are very close to
those of the four QE system, suggesting that description in terms of QP excitations
interacting via VQP (L 2 ) is reasonable.
Sitko et al. [26] introduced a very simple CF hierarchy picture in an attempt to under-
stand Haldane’s hierarchy of Laughlin correlated daughter states and Jain’s sequence
of IQL states with integrally filled CF LLs. Jain’s MFCF picture neglected interac-
1.4 The Composite Fermion Hierarchy 7
Fig. 1.1 The spectra of ten electrons in the lowest Landau level calculated on a Haldane sphere
with 2Q from 25 to 29. The open circles and solid lines mark the lowest energy bands with the
fewest composite fermion quasiparticles. (See [18], for example, and references therein)
tions between QPs. The gaps causing incompressibility were energy separations
between the filled and lowest empty single particle CF LLs. Not all odd denominator
fractions occurred in the Jain sequence ν = n (2 pn ± 1)−1 , where n and p are non-
negative integers. The missing IQL states were ones with partially filled CF QP shells.
The energy gap causing their incompressibility resulted from residual interactions
between the CF QPs. For an initial electron filling factor ν0 , the relation between ν0
−1
and ν0∗ , the effective CF filling factor, satisfied ν0−1 = ν0∗ + 2 p0 , and gave rise to
the Jain states when ν0∗ was equal to an integer n. What happens if ν0∗ is not an integer?
It was suggested [26] that then one could write ν0∗ = n 1 + ν1 , where n 1 was an integer
and ν1 represented the filling factor of the partially filled CF QP shell. If Haldane’s
assumption that the pair interaction energy VQP (L 2 ), as a function of the angular
momentum L 2 of the QP pair, was sufficiently similar to V0 (L 2 ), the interaction
8 1 Fermion Correlations
Fig. 1.2 Energy spectra for N = 12 electrons in LL0 with 2l = 29, and for N = 4 QEs in CF LL1
with 2l = 9. The energy scales are the same, but the QE spectrum was determined using VQE (R)
as the pair psuedopotential (up to an arbitrary constant). (See [18])
energy of the electrons in the LL0, then one could reapply the CF transformation to
−1
the CF QPs by writing ν1∗ = ν1−1 − 2 p1 . Here, ν1 is the CF QP filling factor and
2 p1 is the number of CS flux quanta added to the original CF QPs to produce a second
generation of CFs. For ν1∗ = n 2 , an integer, this results in ν1 = n 2 (2 p1 n 2 ± 1)−1 ,
−1
and a daughter IQL state at ν0−1 = 2 p1 + n 1 + n 2 (2 p1 n 2 + 1)−1 . This new odd
denominator fraction does not belong to the Jain sequence. If ν1∗ is not an integer,
then set ν1∗ = n 2 + ν2 and reapply the CF transformation to the CF QE in the new QP
shell of filling factor ν2 . In general, one finds at the lth generation of the CF hierarchy,
that this procedure generates Haldane’s continued fraction leading to IQL states at
all odd denominator fractional electron fillings. The Jain sequence is a special case
in which ν0∗ = n gives an integral filling of the first CF QP shell, and the gap is the
separation between the last filled and first empty CF levels.
The CF hierarchy picture was tested by Sitko et al. for the simple case of (2l, N ) =
(18, 8) for LL0 by comparing its prediction to the result obtained
through
exact
numerical diagonalization. For this case, 2l0∗ = 2l − 2 N − 1 = 18 − 2 7 = 4.
Therefore, CF LL0 can accommodate 2l0∗ + 1 = 5 CFs. The three remaining CFs
must go into CF LL1 as CF QEs of angular momentum lQE = l0∗ + 1 = 3. This gen-
erates a band of states with L = 0 ⊕ 2 ⊕ 3 ⊕ 4 ⊕ 6. This is exactly what is found
for the lowest energy band of states obtained by numerical diagonalization shown in
Fig. 1.3. Reapplying theCF transformation
to the first generation of CF QEs would
generate 2l1∗ = 2l0∗ − 2 n QE − 1 = 4 − 2 (2) = 0, giving an L = 0 daughter IQL
state if the CF hierarchy were correct. Clearly, the lowest energy state obtained in the
numerical diagonalization does not have angular momentum L = 0 as predicted by
the CF hierarchy. The L = 0 and L = 3 multiplets clearly have higher energies than
the other three multiplets. Sitko et al. conjectured that this must have resulted because
the psuedopotential VQE (L 2 ) was not sufficiently similar to that of electrons in LL0
1.4 The Composite Fermion Hierarchy 9
Pan et al. [27] found IQL states of electrons in LL0 that do not belong to the Jain
sequence of integrally filled CF states. One example is the ν = 4 /11 filling factor of
a state that is assumed to be fully spin polarized. Numerical diagonalization studies
of fully spin polarized systems did not find an L = 0 IQL ground state at νQE =
1
/3 , which would result in an IQL state at electron filling factor of ν = 4 /11 . In
addition, Pan et al. found strong minima in ρx x at even denominator filling factors
10 1 Fermion Correlations
Fig. 1.4 The pseudopotentials of a pair of quasielectrons (left) and quasiholes (right) in Laughlin
ν = 1 /3 (top) and ν = 1 /5 (bottom) states, as a function of relative angular momentum R. Different
symbols mark data obtained in the diagonalization of from 6 to 11 electrons
(ν = 3 /8 and ν = 3 /10 ) suggesting the existence of IQL states that can’t be part
of the CF hierarchy. Our research group has made an important contribution to
this field, by rigorously proving [28–32] under which conditions Jain’s elegant CF
approach correctly predicts the angular momentum multiplets belonging to the lowest
energy sector of the spectrum for any value of the applied magnetic field. Because
there is no small parameter in this strongly interacting many-body system, our proof
does not involve treating fluctuations beyond the MF by a perturbation expansion. It
involves proving some rigorous mathematical theorems and applying them, together
with well-known concepts frequently used in atomic and nuclear physics. We do
not review the arguments on why and when Jain’s MFCF picture correctly predicts
the angular momentum multiplets in the lowest band but urge the interested reader
to look at [28–32]. Our rigorous theorem that should be emphasized states that
for
a harmonic
pair psuedopotential VH (L 2 ) = A + B L 2 (L 2 + 1), every multiplet
N
l , Lα with the same value of total angular momentum L has the same energy,
1
E α (L) = N (N − 1) A + B (N − 2) l (l + 1)
2
+ B L (L + 1) , (1.3)
Fig. 1.5 Pair interaction pseudopotentials as a function of relative angular momentum R2 for
electrons in the LL0 (a), LL1 (b) and for the QEs of the Laughlin ν = 1 /3 state by Lee et al. [33,
34] (squares), and by Wójs et al. [29, 30] (triangles). (See [18])
tiplets having the same value of total angular momentum L). Only the deviation
V (L 2 ) defined by V (L 2 ) = V (L 2 ) − VH (L 2 ) results in correlations. The sim-
plest model for V (L 2 ) is one with only short range anharmonic behavior,
V (L 2 ) = kδ (L 2 , 2l − 1) . (1.4)
If k > 0, it is apparent that the lowest energy multiplet for each value of total angular
momentum L is the one which avoids (to the maximum possible extent) having pairs
with L 2 = L MAX2 = 2l − 1. This is just what is meant by Laughlin correlations, and
is the reason why the Laughlin trial wave function for the ν = 1 /3 filled state is the
exact solution to a short range pair interaction psuedopotential. If k < 0, then the
lowest energy state (when ν is not too small) for each value of L is the one with
MAX N
PLα L 2 , the probability that l , Lα has pairs with pair angular momentum
L 2 = L MAX
2 , having a maximum value. This corresponds to forming pairs with L 2 =
L MAX
2 = 2l − 1. It is important to emphasize that V (L 2 ) rises faster than VH (L 2 ) at
all values of L 2 for electrons in LL0. In LL1, this is not true for 2l − 1 ≥ L 2 ≥ 2l − 5.
Thus in LL1, we do not expect Laughlin correlations for filling factor ν = 1 /3 .
In Fig. 1.5, we display V (R2 ), where R2 = 2l − L 2 for a) electrons in LL0,
b) electrons in LL1, and c) QEs in CFLL1. It is clear that for QEs of the ν =
1
/3 filled IQL state, VQ E (L 2 ) is not increasing with L 2 faster than the harmonic
psuedopotential. In fact, unlike V0 (L 2 ) and V1 (L 2 ), it is not even a monotonically
increasing function of L 2 . Therefore, we certainly do not expect Laughlin correlations
among the CFQEs in CFLL1.
Moore and Read [35] treated the ν = 5 /2 state (thought of as ν = 2 + 1 /2 ) as a
half filled spin polarized state of LL1 with both spin states of LL0 occupied. They
suggested that the state had “pairing correlations ” similar to those occurring in
12 1 Fermion Correlations
Fig. 1.6 The pseudopotentials VQE (R2 ) and VQH (R2 ) for a QEs of ν = 1 /3 state, b QHs of ν = 1 /3
state, and c QHs of ν = 2 /5 state. (See [19])
For a single pair, lFP = 2l − 1. As NP increases, the allowed values of the total
angular momentum are restricted to values less than or equal to 2lFP . The value of
the constant γF is determined by requiring that the FP filling factor be equal to unity
when the single Fermion filling factor has an appropriate value. For lP = 2l − 1, this
−1
value corresponds to single Fermion filling ν = 1. Setting νFP = (2lFP + 1) /NP ,
−1
ν −1 = (2l + 1) /N , and NP = N /2 gives νFP = 4ν −1 − 3, (i.e., γF = 3), so that
νFP = 1 when ν = 1. The factor of 4 multiplying ν −1 results from the pairs having a
charge of −2e, and NP being equal to N /2. This procedure allows the Fermion pairs
to be Laughlin correlated instead of the individual electrons being so. It predicts that
an IQL state at ν1 = 1 /2 occurs when 2l = 2N − 3 as found by Moore and Read.
We have applied the same idea to the quasielectrons and quasiholes of the Laughlin
ν = 1 /3 state, and to quasiholes of the Laughlin–Jain ν = 2 /5 state. QHs of ν = 1 /3
state reside in CF LL0, but both QEs of the ν = 1 /3 state and the QHs of the ν = 2 /5
state reside in CF LL1.
If we assume that the QEs form pairs and treat the pairs as Fermions, then (1.5)
gives the relation between the effective FP angular momentum lFP , and the QE angu-
1.5 Justification of the CF Approach 13
lar momentum l, and the relation between the effective FP filling factor νFP , and
the QE filling factor νQE . If we take νFP = m −1 , when m is an odd integer, we can
obtain the value of νQE corresponding to the Laughlin correlated state of FPs (pairs
of quasielectrons with lP = 2l − 1). Exactly the same procedure can be applied to
QHs in CF LL1 since VQE (R) and VQH (R) are qualitatively similar at small values
of R. Here, we are assuming that VQE (R) and VQH (R) are dominated by their short
range behavior R ≤ 5. The QH psuedopotential is not as well determined for R > 5
because it requires larger N electron systems than we can treat numerically. The elec-
−1 −1
tron filling factor is given by ν −1 = 2 + 1 + νQE or by ν −1 = 2 + 2 − νQH .
This results in the values of ν in qualitative agreement with experimental results.
Confirmation of the Laughlin explanation of when IQL ground states occur can
be found through numerical diagonalization of the Coulomb interaction between
electrons. For ωc large compared to the Coulomb energy scale Vc e2 /λ, only
the subspace of the lowest Landau level (LL0) is relevant to determining the low
energy states in the spectrum. Numerical diagonalization is usually performed in
Haldane’s spherical geometry. A concise explanation of this set-up is as follows.
There is a one-to-one correspondence between N electrons in a plane described by
coordinates (r, φ) and the N electrons on a sphere described by (l, l z ). For the plane,
the z-component of angular momentum takes on the values m = 0, 1, . . . , Nφ and
N
the total z-component of angular momentum is M = i=1 m i , where m i is the z-
component of the angular momentum of particle i. M is the sum of the relative angular
momentum M R and the center of mass angular momentum MC M . On a sphere, the
z-component of the single particle angular momentum is written as l z , and |l z | ≤ l,
where l is the angular momentum in the shell (or Landau level). The total angular
momentum L is the sum of the angular momenta of N Fermions, each with angular
momentum l. N electron states are designated by |L , L z , α α, where α is used to
label different multiplets with the same value L. It is apparent that M = Nl + L z
and one can show that M R = Nl − L and MC M = L + L z . Therefore, for a state of
angular momentum L = 0, M R must be equal to Nl. In general, the value of L for
a given correlation function G is determined by the equation L = Nl − K F − K G ,
where K F = N (N − 1)/2 is the number of CF-lines appearing in the Fermi function
F and K G is the number of CF-lines appearing in the correlation function G. For
L = 0, this relation was given by [36]. In Haldane’s spherical geometry, the N
electron system is confined to move on the surface of a sphere of radius R, with a
magnetic monopole of strength 2Q flux quanta sitting at the center (2Q is taken to be
an integer). This results in a radial magnetic field of magnitude B0 = 2Qφ0 /4π R 2 .
This geometry avoids boundary conditions, has full rotational symmetry, and contains
a finite number of particles. The single particle
eigenfunctions are called monopole
harmonics [37] and can be written as Qlm . They are eigenfunctions of the single
14 1 Fermion Correlations
and the reduced matrix element on the right hand side is independent of M. The
matrices to be diagonalized are very large but sparse, and standard programs allow
diagonalization for N up to roughly 12 or 14. The numerical results displayed in
Figs. 1.1 and 1.2 are representative of the general results obtained for N ≤ 14. It
should be noted in Fig. 1.1 that there is a gap between the lowest state or band of
states and a quasi continuum above them. It should also be stressed that the spread in
values of energy for the 2QP bands (frames d and e) is small compared to this gap.
A reasonable understanding of the ground states and elementary excitations of
the states in the Laughlin–Jain sequence for partially filled LL0 results from the
agreement of the Laughlin–Jain correlations with the numerical results obtained by
diagonalization within the partially occupied Landau level.
Laughlin [15] realized that if the interacting electrons could avoid the most strongly
repulsive pair states, an incompressible quantum liquid (IQL) state could result. He
suggested a trial wave function for a filling factor ν equal
to the reciprocal
of an odd
integer n, in which the correlation function G n z i j was given by i< j z in−1 j . This
function is symmetric and avoids all pair states with relative pair angular momentum
smaller than n (or all pair separations smaller than rn = n 1/2 λ). One can represent
this Laughlin correlation function G L diagrammatically by distributing N dots, rep-
resenting N electrons on the circumference of a circle, and drawing double lines,
representing two correlation factors (cfs) connecting each pair. There are 2 (N − 1)
1.7 Correlations and Correlation Diagrams 15
CF factors in G L z i j emanating from each particle i. Adding (N− 1) CF fac-
tors emanating from each particle due to the Fermion factor F z i j gives a total
of 3 (N − 1) cfs emanating from each particle in the trial wave function . This
number cannot exceed Nφ = 2l defining the function space (2l, N ) of the LL0.
The other well-known trial wave function is the Moore–Read [35] paired function
describing the IQL state of a half filled spin polarized first excited Landau level
(LL1). This wave function can be written in the form = F · G MR , where the
−1
correlation function is taken as G MR = F z i j P f z i j . The second factor is called
the Pfaffian of z i−1
j . It can be expressed as [25, 35]
N /2
Pf z i−1
j = Â −1
(z 2i−1 − z 2i ) , (1.7)
i=1
Fig. 1.7 Moore–Read correlation diagram for N = 4. Dots represent the particles, and solid lines
the cfs z i j . G MR is the symmetric sum of the three diagrams and is given by (1.9)
Fig. 1.8 Quadratic correlation functions. The square of a CF, z i2j , is represented by double lines.
G Q is the sum of the contributions from the three diagrams
by our group before we discovered that Cappelli et al. [39] had already shown the
equivalence.
There are several advantages to the use of G Q . First, it is simpler to parti-
tion N into two subsets A and B of size N /2, (e.g., A = {1, 2, . . . , 2N /2} and
B = {N /2 + 1, . . . , N }), and define g AB = g A g B = i< j∈A z i2j k<l∈B z kl for each
(A, B). Then the full correlation function can be written as Ŝ N {g AB }, where Ŝ N
symmetrizes g AB over all N particles. This symmetrization is equivalent to sum-
ming g AB over all possible partitions of N into two equal size subsets A and B. In
Fig. 1.9, we show the contribution to G Q for N = 8 particles for a partition in which
A = {1, 3, 5, 7} and B = {2, 4, 6, 8}.
Jain [16, 17] introduced a composite Fermion (CF) picture by attaching to each
electron (via a gauge transformation) a flux tube which carried an even number 2 p
of magnetic flux quanta. This Chern–Simons (CS) flux has no effect on the classical
equations of motion since the CS magnetic field b (r) = 2 pφ0 i δ (r − r i ) ẑ van-
ishes at the position of each electron (it is assumed that no electron senses its own
CS flux). Here, φ0 is the quantum of flux, and the sum is over all electron coordinates
r i . The classical Lorentz force on the ith electron due to the CS magnetic field is
(−e/c) v i × b (r i ), and b (r i ) caused by the CS flux on every j not equal to i van-
ishes at the position r i . The CF model results in a much more complicated interaction
Hamiltonian, but simplification results from making a mean field (MF) approxima-
tion in which the CS flux and the electron charge are uniformly distributed over the
1.7 Correlations and Correlation Diagrams 17
entire sample. The average electronic charge −eN /A is canceled by the fixed back-
ground of positive charge introduced to make the total charge vanish. This MFCF
approximation results in a system of N non-interacting CFs (CF = electron plus
attached flux tube) moving in an effective magnetic field b∗ = νb. An effective CF
filling factor ν ∗ was introduced satisfying the equation
∗ −1
ν = ν −1 − 2 p. (1.10)
Table 1.2 Values of l for an N = 4 electron system and the values of l0∗ , n QE , lQE , k M , and L which
result
l l0∗ n QE lQE kM L
4.5 1.5 0 2.5 6 0
4 1 1 2 5 2
3.5 0.5 2 1.5 4 0⊕2
3 0 3 1 3 0
Mutta kun hän tuli alas, ei hän löytänyt äitinsä tupaa. Se ei ollut
entisellä paikallaan, — portailla ei kasvanut sammalta, eikä niiden
ympärillä kukkia.
Silloin tuli vanha mies ja otti häntä kädestä. — Näen että olet tullut
takaisin! Kiltisti, että läksit tänne katsomaan!
— Entä Dordi?
Hänen täytyi istua suurelle kivelle, joka oli ollut oven edessä.
Hän siveli sitä käsillänsä. — Vai olet sinä sentään jäljellä! Sinä
kaiketi vielä tunnet minut?
Auringon laskiessa istui hän siinä vielä. Ja kun se nousi, istui hän
siinä yhä edelleen.
— Sinun täytyy sallia minun istua, olen niin väsynyt! Tuli ilta, ja
hän näki auringon laskevan.
*****
*****
*****
*****
— Tunsin kerran miehen, jota onni ei ollut seurannut. Hän oli luotu
elämään suurissa oloissa ja joutui pieniin. Pieni kaupunki kiusasi
häntä, hänen työnsä kiusasi häntä. Niin tuli hän kesällä tänne
muutamaksi ajaksi. Hän kartteli ihmisiä. Täällä ylhäällä hän oli
iloinen. Täällä hän voi laskea leikkiä ja nauraa. Täällä voi hän
käyttää suuria mittoja, sanoi hän. Hän käyskenteli täällä valkoisten
huippujen keskellä, aina yksin. Päivää ennen kuin hänen oli jälleen
matkustettava takaisin kouluunsa, kulki hän liian kauas — eikä
palannut enää koskaan.
— Ei, elämänkokemus.
— Tietysti.
— Oli kesä. Hän käyskenteli yksin. Näytti siltä kuin hän kuulisi
kaikki etäisyydestä vain.
— Niityt kutsuelivat häntä. Oli metsäkukkasten aika.
*****
— Oli mies, joka oli tuntenut hänet — kauan sitten. Kun hän näki
hänet jälleen sellaisena, vavahti hänen sydämensä. Hän olisi
halunnut langeta hänen jalkojensa juureen.
— Silloin tuli hänen luoksensa tuo mies, joka oli tuntenut hänet
muinoin. Hän poimi kukkia ja tahtoi antaa ne hänelle. Mutta Sanpriel
kääntyi hänestä pois.
*****
Jos Samuel Stern halusi laskea leikkiä, miksi ei hän yhtyisi siihen!
Hän alkoi puhua, mutta hänen äänensä vapisi niin ettei hän heti
saanut sanoja esiin. Mutta pian se vakiintui.
— Ah, itse sinä olit syypää siihen että minun täyt mennä!… Sinun
syysi oli, että jouduin sinne, minne minun ei olisi pitänyt joutua, —
että jouduin liiaksi pois suunniltani! Olen vaeltanut pimeydessä. — —
—